Bryan Caplan  

Meritocracy Bleg

The problem with "shocks"... Dean Baker on Something "All E...
If every worker were paid exactly proportional to their actual productivity - if neither statistical discrimination nor fairness played any role in the job market - how much would the standard deviation of labor earnings increase?

HT: Garett Jones

COMMENTS (18 to date)
Effem writes:

How do you define productivity? Is an illegal drug dealer productive?

dangerman writes:

Economy wide: 10x

In certain industries: 1,000x +

Standards regarding group cohesion and social standing would never allow this to happen.

Michael B. writes:

If all workers were paid exactly proportional to actual productivity, then employee compensation would be fair. Unfortunately, there's no way to measure productivity in this way. What metrics that exist can be fooled. Take the example of a grocery stocker because such low level positions cannot negotiate wages the way that skilled labor can. They can stock things faster if they indiscriminately shove things onto the shelves, mixing up or hiding other merchandise. Management would be none the wiser and no metric would detect this. It might hurt sales though as each product would not have the optimum amount of merchandising space and products people might want would be hidden.

love actuary writes:

A negative amount - the standard deviation might decrease.

If we assume that many salaries are disproportionately higher than worker productivity (for instance, because of trade restrictions, political-regulatory influence of the company, ect), then aligning earnings with productivity may tighten the distribution and lower the standard deviation of earnings.

If "worker(s) were paid exactly proportional to their actual productivity" implies that certain sectors do not receive the political benefit of regulation or the potential bail-out, the distribution of earnings may tighten around the mean and not expand. The distribution of earnings may have something like a median of 40-50k and a higher mean due to a $0 lower bound.

JJ writes:

In this hypothetical, does the minimum wage still exist? How do we account for the many low skilled workers who provide negative value? What about the CEO who through incompetence costs his company millions of dollars?

How are we measuring productivity? Is it through a process comparing the worker to a replacement level worker?

Fazal Majid writes:

The generally accepted difference in productivity between a star programmer and a merely average one is 10x. This is considered specially noteworthy and thus the average across all industries is probably far less than that. The question is, how many workers are poor? If you believe Sturgeon's law, 80%, but that seems rather high, I would guess 30% or so. Thus the distribution would be 30% of employees with productivity 0, 69% with productivity 1 and 1% with productivity 10, for a standard deviation of roughly 1x the unit productivity, or 1.3x mean productivity.

Of course, the whole premise of the question is that productivity is an individual characteristic, and completely discounts the value of teamwork, or the management that fosters it or not. In most cases, teamwork dominates individual productivity, after all, if it did not, there would not be any point in having firms as opposed to networks of individual contractors.

Ben writes:

My hunch is that it would decrease. There are doubtless differences in productivity among workers who are presently paid the same amount; but my guess is that that is dwarfed by the high salaries paid to the many administrators and management who contribute little, if anything, to productivity. (Which is not to say that all administrators and managers are leeches; only that many are.)

Granite26 writes:

Are you assuming that you can perfectly measure productivity (via magic?) without affecting that productivity?

Less snarky, is this a theoretical question about distribution today, or a theoretical question about what would happen if measurement got better and marginal product could be effectively measured and distributed?

Floccina writes:

I do not know how to quantify it to answer the question but back when I managed restaurants I noticed that some dishwashers would do more that 2x the work of others and would be paid a little more, (and might be promoted to job that they would less good at.)

Now in software, where I work now, I would say that some coders and do 5x more the work of others and they might get double the pay of the others.

john hare writes:

There would be a significant closing of the spread.
Some extreme earners would still get the megabucks while others would lose. On the other end though, low earners would learn to be more productive or starve.

Anybody that has watched people on percentage or piecework compared to people doing theoretically the same job will notice a 2-5 times difference in average productivity. Even restaurants show this effect with waitresses that split tips requiring considerably more staff than those where the individuals keep their own.

So if you had an accurate magic tech that measured and paid accordingly, GDP would double in five years.

Miguel Madeira writes:

By productivity you are talking about average productivity or marginal productivity?

Toby writes:

It depends I suppose... for example:

1. Fairness (probably) decreases the variance of income;

2. Compensation structures to set efficient levels of effort when productivity cannot be observed, such as tournaments, increase the variance of income.

Whether #1 or #2 dominates is not something I can answer let alone quantify.

I am willing to hazard a guess though that ability is normally distributed -- IQ is normally distributed and this seems to be a good enough proxy for ability -- and that the standard deviation of labor earnings will thus decrease as I believe labor earnings are probably distributed according to something such as a Pareto distribution. I am also guessing that mean income would probably increase by between 25% and 50%... as this seems to be at most the expenditure that is currently devoted to measure productivity. It's a wild wild guess though and most likely wrong.

Glen Smith writes:

My first question is about how productivity is measured here. If by some magic one could figure out real productivity as opposed to perceived productivity, my hunch would be that standard deviation in income distribution would decline. Average income would probably decline relative to median income.

BC writes:

One might gain some clues by looking at the variance of earnings among professional athletes, taking into account that many aspiring athletes essentially earn zero. Sports labor markets have a lot of information symmetry since teams can all observe each others players' productivity, are very transparent since athletes' compensation is usually publicized, and are very liquid since players change teams and negotiate new contracts quite often. All these factors suggest that athletes' pay is highly "merit based".

Side note: one definition of "fairness", and probably the most reasonable definition at that, is precisely that every worker is paid exactly proportional to productivity. So, it's strange to describe the premise of the question as one where fairness plays no role in the job market.

ThomasH writes:

Or decrease?

Yavor writes:

It will likely increase a lot provided we consider "inclusive productivity" - in other words some people's productivity creating cascading effects on multiple other people's productivity. Presumably, this will be a logarithmic increase....winner take all type of thing.

As an aside - a lot of jobs may not be technically valuable due to productivity. For example, a regulator's position may be more concerned with organizational stability/security which may or may not be conducive to productivity estimation - no way to estimate the counterfactual.

Brian M. Saxton writes:

Two questions:

1. As a couple of people have already mentioned, the productivity of any resource (including human capital) is highly dependent on the presence of complimentary assets. So, when we talk about paying on an individual productivity basis, are we talking the individual's productivity in their current setting, or their productivity with a "generic" set of assets, or what? (This issue is pretty important with respect to human capital in particular because its productivity is so often dependent on co-specialized assets. I'm fairly productive in the classroom, as long as someone else is recruiting and billing students. Without a business office to make sure students pay, I'm just shouting about strategy on the street corner and I'm pretty sure that's negatively productive.)

2. I assume that the condition of perfect fairness is supposed to abstract away from Alchian and Demsetz (1972) measurement of team production issues, and also abstracting away from Beckerian (1964) questions of value appropriation, but those issues become important pretty quickly.

Also, I was at your talk in Bowling Green a couple of weeks ago. Thanks; it was quite thought provoking and your thesis led to some interesting class discussion last week.

Roger writes:


I am simply amazed by the answers above.Most of those that aren't evasive seem spectacularly wrong. IMHO.

The answer is that average wages would skyrocket. Those who are extremely productive would be rewarded with substantially higher multiples of pay.

Those who are less productive would see their pay plummet unless they changed their behavior. They would clearly see what the productive ones are doing and would imitate them. If they were unable to reform their habits and productivity, economic incentives would dictate that they change their line of work. Thus they would seek out assignments and techniques which allow them to optimize their productivity.

The increase in productivity would lead to fewer people doing each job, which would free people to do additional value added jobs serving their fellow man. We would thus see a virtuous cycle of unimaginable increases in human prosperity and well being and higher standards of living across the board.

This is true even for the incorrigibly less productive by the way, due to the concept of comparative advantage. Once the productive did what they are most productive at, the net benefit of paying barely productive people would rise too -- substantially (gardeners and barbers much make more in wealthy nations).

As to people who are net intrinsically unproductive at everything, they would not be employed. They could however reap the transfer benefits of the higher productivity of everyone else.

The way to look at this is that the thought exercise makes productive organizations substantially more efficient. Efficient production is efficient problem solving, and cumulative problem solving is what our prosperity hinges upon.

This imaginary world would have no free riding within organizations. Everyone would pull their own weight and be rewarded for it (I assume the perfect productivity measures include perfect measures of net contributions to complex teams).

Granted I can imagine all kinds of problems with transitioning from today's imperfect productivity to one of omniscient productivity. But if we take the thought exercise on at face value, the answer is pretty clear.

I would add that the exercise assumes proportionality to productivity. Earnings would still be set by supply and demand, it is just that the ranges would be productivity proportionate*

Indeed the question I have is for all the commenters above who supposedly understand economics (they read this blog) yet come to the opposite conclusion. How can you explain your answers and/or why am I wrong?

Just asking, I am well aware that I could be wrong and I would love to know why.

* as an executive at a major corporation my contributions (or economic harm) to the firm were several orders of magnitude higher than my ample salary and bonus. Still, I would have done the job for half my salary, but supply and demand and the threat of me going elsewhere led to the actual salary paid. This would still exist in the world of perfect profuctivity transparency and reward.

Comments for this entry have been closed
Return to top